81 / 97
Risk of Ruin
Working definition
The probability that an account's equity reaches a barrier from which it cannot continue — margin exhaustion, a drawdown limit, or zero — before the strategy's edge has time to assert itself.
Risk of ruin is the oldest result in gambling mathematics pointed at a trading account. The classical gambler’s-ruin setup asks: given an edge per bet, a stake size, and a finite bankroll, what is the probability of hitting zero before running indefinitely? The answers carry three lessons that survive every refinement. With a negative edge, ruin is certain given enough play. With a positive edge, ruin is still strictly positive — an edge is not immunity, only a tendency. And the probability is savagely nonlinear in stake size: betting twice as large does not double the risk of ruin, it can raise it by orders of magnitude, because ruin is driven by the tail of loss sequences and the tail compounds.
Textbook treatments then observe that fixed-fractional sizing drives classical ruin toward zero — a fraction of a shrinking bankroll can never quite reach nothing. Real accounts are not so lucky, because real ruin barriers sit far above zero: the margin call, the prop firm’s daily loss limit, the trailing drawdown floor, the investor’s redemption trigger, the trader’s own capitulation. Ruin is properly defined by the nearest binding barrier, and against those barriers fractional sizing offers no asymptotic escape. Fat-tailed returns and correlated positions — five trades that are secretly one trade — raise the probability further beyond what Gaussian intuition suggests.
The statistic’s honest use is conditional. Given a sizing rule, a barrier, and an estimated edge, risk of ruin is computable, and the computation is genuinely decision-relevant: it converts “how much should I risk?” into an explicit survival probability. But it is conditioned on the edge estimate, and edge estimates inherited from an overfit backtest are inflated — which means the computed ruin probability is understated at exactly the moment it is being relied upon. A ruin calculation is only as honest as the drawdown and return distributions fed into it.
The risk of ruin calculator computes it from your own win rate, payoff ratio and stake — and, because a win rate is an estimate from a finite sample rather than a property you can read off, it also asks how many trades yours came from and reports the range that sample supports. The answer is violently non-linear in the win rate, so a few points of sampling uncertainty can move it by an order of magnitude; a single figure hides exactly that.
Two neighbouring questions have their own arithmetic. What stake a record actually supports is the Kelly criterion calculator, which reports the full and fractional stake with the range the win-rate sample supports beside them — a record too short to rule out an edge of zero is a record whose stake at the pessimistic end is nothing, and that is the useful half of the answer. What a chosen risk fraction becomes in lots, once a stop distance, a pip value and the cost of the round trip are named, is the position size calculator.
Commonly confused with
Neighbouring concepts that get used interchangeably, and the distinction that actually separates them.
- Drawdown
A drawdown is an excursion below the high-water mark; risk of ruin is the probability that such an excursion reaches a barrier it cannot come back from. Drawdown describes the path. Ruin describes the path meeting a wall, and the wall is not part of the strategy.
- Maximum drawdown
Maximum drawdown is the deepest excursion you have already observed — a backward-looking description of one realised path. Risk of ruin is a forward-looking probability over the paths you have not walked yet. A record with a modest maximum drawdown can carry a substantial ruin probability.
- Trailing drawdown
A trailing rule makes the ruin barrier move upward with the high-water mark the rule samples — equity including open trades under an intraday rule, a closed daily balance under an end-of-day one — so gains permanently raise the level at which you are terminated. It is the same concept with a barrier that will not stay still, and it defeats the intuition that being ahead makes you safer.
- Position sizing
Sizing is an input to the calculation, not a synonym for it. The same edge sized two ways produces two ruin probabilities, and the relationship is savagely nonlinear — which is precisely why the calculation is worth doing rather than assuming.
How to measure it in your own data
A definition you cannot test is a definition you have to take on trust. This is the shortest honest route from the concept to a number you computed yourself.
- Records you need
A sizing rule, the nearest binding barrier, and an estimated edge with the return distribution behind it. Naming the barrier honestly is most of the work: the margin call, the prop firm's daily loss limit, the trailing floor, the investor's redemption trigger, and your own capitulation point are all candidates, and the binding one is whichever sits closest.
- What you compute
The probability that equity reaches that barrier before the edge has time to assert itself, computed against your actual return distribution rather than a Gaussian one — and with correlation accounted for, because five positions that are secretly one position do not diversify anything.
- What the answer tells you
Three results survive every refinement of the classical setup. With a negative edge, ruin is certain given enough play. With a positive edge, ruin is still strictly positive — an edge is a tendency, not immunity. And the probability is savagely nonlinear in stake size, because ruin is driven by the tail of loss sequences and the tail compounds. The weak input is always the edge estimate: one inherited from an overfit backtest is inflated, so the computed ruin probability is understated at exactly the moment it is being relied upon.
If this has already cost you
If an account has already crossed a barrier, the sizing that carried it there can be measured against the rules it was running under.
- Prop-Breach Forensics“Why did my prop evaluation actually fail?”Will not establish: Whether the firm’s rules are fair, or whether you would have passed with different luck. It reconstructs what happened; it does not adjudicate the firm.
- Stress Replay Assay“What happens to my sizing when the market breaks?”Will not establish: The next shock. By construction the events that break books are the ones the sample did not contain, and the report carries that sentence rather than burying it.
Intake is not open yet, so none of these can be commissioned today. They are listed here so you know the measurement exists and what it would and would not settle — the launch list hears first.
Work it out yourself
Free calculators that take this concept as an input. Each shows its working, so the number it gives you can be checked rather than taken on trust.
Questions and answers
Does having a positive edge mean I cannot be ruined?
No. With a positive edge the probability of ruin is lower, and it remains strictly positive. An edge is a tendency over many repetitions, not a guard rail on any particular sequence, and the sequence is what ends accounts. The classical gambler's-ruin result makes this precise and it does not soften with sophistication.
Why does doubling my position size more than double my risk?
Because ruin is driven by the tail of loss sequences rather than by average outcomes, and tails compound. Doubling the stake does not double the probability of an unlucky run reaching the barrier — it can raise it by orders of magnitude. This nonlinearity is the single most useful thing the statistic tells a trader.
Where is my actual ruin barrier?
Almost certainly far above zero, which is where textbook treatments put it. Real barriers bind first: margin exhaustion, a prop firm's daily loss limit, a trailing drawdown floor, an investor's redemption trigger, or the point at which you personally stop. Fixed-fractional sizing drives classical ruin toward zero, and offers no such escape against a barrier that sits above it.
Why might my computed risk of ruin be too optimistic?
Because it is conditioned on an edge estimate, and edge estimates from a heavily searched backtest are inflated. Fat tails and correlated positions push the true probability further above what Gaussian intuition suggests. A ruin calculation is only as honest as the return and drawdown distributions fed into it.
Related terms
Derived from the links this entry makes and the entries that link back to it.
In the research
Risk of Ruin comes up in five research notes on this site, and this entry lists three of them.